A curve fitting methodology utilizing the Levenberg-Marquardt method for fitting non-linear models was extended to norms other than L (least squares). Heuristics are proposed and programs have been developed which offer these methods as practical tools. Manuscripts describing these methods have been submitted for publication. A second project area involved a comparison of methods for organizing very large amounts of stored data which facilitate rapid storage and retrieval. Existing hashing methods were evaluated and new algorithms are proposed for very large data bases. A methodology for evaluation and comparison of schemes is described.